Noise maps provide city-scale perception of noise levels. However, existing mapping systems suffer from several limitations: They mostly rely on historical data, lack precise noise-source contextual information, and perform audio processing in the cloud, creating only near real-time noise maps. In this work, we explore whether audio sensing and processing can both be performed on edge devices, enabling the creation of dynamic noise maps with real-time contextual visualization. The core challenge is to design a lightweight audio classification model that runs on edge devices without compromising performance. An end-to-end prototype system has been designed for detecting vehicular honks and their subclassification in terms of vehicle type. This work designs a deep learning model, GridCNN, that takes spectrogram images of 1-sec audio samples and outperforms existing audio classification algorithms (13.23% higher accuracy) using 63.14% fewer parameters and consuming 84.21% less memory. The model deployed on a Raspberry Pi Zero 2 W edge device using TensorFlow Lite achieves 94.62% test accuracy. Suitable real-time visualization of honk information has been created on maps. For model training, suitable honk datasets (56290 samples) were prepared by collecting and annotating in-house audio data using a smartphone sensor under different traffic scenarios. Honk-related contextual details can yield valuable insights to help municipal authorities devise appropriate traffic planning policies to curb noise pollution. The proposed GridCNN model may be generalized through suitable tuning and training to enable real-time classification of other outdoor noise sources (engine noise, firecrackers, loudspeakers, etc.) as well as indoor noise classes on edge devices.

Driver drowsiness is one of the leading causes of road accidents worldwide, particularly in resource-constrained and high-variability environments where real-time monitoring is critical. To address this challenge, we present DRISTI (Driver Real-time Intelligent Spatio-Temporal Inference), a novel deep learning framework designed for efficient and accurate drowsiness detection. Unlike traditional convolution-based systems, DRISTI leverages a Vision Transformer (ViT) encoder to extract global spatial representations from facial frames, enabling the capture of fine-grained cues such as eye closure, yawning, and gaze shifts with reduced computational cost. These embeddings are modelled through a Long Short-Term Memory (LSTM) network to learn temporal behavioural dynamics, while an additive attention mechanism highlights the most informative segments, ensuring reliable recognition of both transient and sustained drowsiness. A dynamic frame selection strategy further reduces redundancy by retaining only behaviourally distinct frames, improving efficiency without compromising accuracy. DRISTI attains 89.9% Accuracy, 91.2% Macro F1, and 93.5% Weighted F1, with ROC-AUCs of 0.98 (Not Drowsy), 0.96 (Drowsy), and 0.92 (Transient Drowsy). These results indicate robust, class-balanced performance suited for safety-critical early-warning applications.

This research explores the classification of environmental sounds using the UrbanSound8K dataset, introducing a novel feature fusion technique that combines spectrograms, scalograms, and Mel-Frequency Cepstral Coefficients (MFCCs) - a combination not previously explored in existing literature. While individual features such as spectrograms, scalograms, and MFCCs achieved accuracies of 92%, 89%, and 89%, respectively, the fusion of these features led to a significant improvement in accuracy, reaching 97%, with a notable 5-8% performance gain. The proposed method harnesses the complementary nature of these features, effectively capturing temporal, frequency, and perceptual characteristics of audio signals, enabling more robust and comprehensive representations. The fused features are processed through an enhanced AlexNet architecture, customized to handle multi-dimensional inputs. The model demonstrated excellent noise robustness, faster convergence, and superior generalizability compared to models trained on individual features. The findings pave the way for future applications, including real-time environmental sound classification integrated with IoT devices, mobile applications, and broader domains such as wildlife monitoring and industrial noise detection systems.

Ensuring safe drinking water is crucial for public health, yet existing assessment methods often fail to capture complex contaminant interactions, provide real-time analysis, or offer model interpretability. Traditional threshold-based evaluations lack adaptability, while many machine learning models are opaque, making it difficult to understand key influencing factors. To address these challenges, we propose SafeH2O, a data-driven framework integrating machine learning and explainability techniques. The system applies data preprocessing, including normalization, feature scaling, and class balancing using SMOTE, to enhance dataset reliability. K-means clustering is employed to categorize water samples, followed by supervised classifiers - such as Random Forest, Logistic Regression, and Gradient Boosting - to predict water potability. Among them, Gradient Boosting achieves the highest accuracy of 93%. To improve interpretability, local interpretable model-agnostic explanations (LIME) is used to analyze key contaminants influencing water quality across clusters. Additionally, we design a health risk assessment microservice that issues real-time contamination alerts and safe water recommendations.

Some recent studies highlight that vehicular traffic and honking contribute to more than 50% of noise pollution in urban or sub-urban areas in developing countries, including Indian cities. Frequent honking has an adverse effect on health and hampers road safety, the environment, etc. Therefore, recognizing the various vehicle honks and classifying the honk of different vehicles can provide good insights into environmental noise pollution. Moreover, classifying honks based on vehicle types allows for the inference of contextual information of a location, area, or traffic. So far, the researchers have done outdoor sound classification and honk detection, where vehicular honks are collected in a controlled environment or in the absence of ambient noise. Such classification models fail to classify honk based on vehicle types. Therefore, it becomes imperative to design a system that can detect and classify honks of different types of vehicles to infer some contextual information. This paper presents a novel framework AClassiHonk that performs raw vehicular honk sensing, data labeling, and classifies the honk into three major groups, i.e., light-weight vehicles, medium-weight vehicles, and heavy-weight vehicles. Raw audio samples of different vehicular honking are collected based on spatio-temporal characteristics and converted them into spectrogram images. A deep learning-based multi-label autoencoder model (MAE) is proposed for automated labeling of the unlabeled data samples, which provides 97.64% accuracy in contrast to existing deep learning-based data labeling methods. Further, various pre-trained models, namely Inception V3, ResNet50, MobileNet, and ShuffleNet are used and proposed an Ensembled Transfer Learning model (EnTL) for vehicle honks classification and performed comparative analysis. Results reveal that EnTL exhibits the best performance compared to pre-trained models and achieves 96.72% accuracy in our dataset. In addition, context of a location is identified based on these classified honk signatures in a city.

Classifying citation trajectories of scientific publications is crucial. However, they diffuse anomalously due to non-linear, non-stationary, and long-ranged correlations. Previous studies define hard thresholds, arbitrary parameters, and subjective rules to classify based on their rise and fall patterns. It leads to substantial variance and, thus, ambiguous classification. This paper proposes CiteDEK, a hybrid EMD-kNN-DTW classification model framework. It predicts the nature of 5,039 trajectories, each 30 years in length, using only raw time series. We get a classification accuracy of ≈ 76%, and Cohen's kappa-statistic is 0.63, which is significant.

Noise pollution from vehicular honking is a pressing issue in urban environments, necessitating the development of efficient honk detection systems. While deep learning models have shown promise in detecting car honks, their high computational demands limit their deployment on mobile devices. This research introduces a novel approach to car honk detection, introducing a lightweight deep learning models optimized for mobile devices. We developed a mobile application that detects vehicular honks in real-time, offering a practical solution to monitor noise pollution and enhance pedestrian safety. Our proposed model achieved a 97.65% accuracy with low average inference time and latency of 63.8 ms and 288.79 ms, respectively. This application can alert pedestrians wearing earphones or headphones about nearby honks, helping prevent accidents. Our work addresses key challenges in resource optimization, real-time processing, and latency reduction, providing a significant contribution to urban safety and environmental monitoring.

Human activity recognition (HAR) has gained significant attention in recent years. Previous studies used various features from time, frequency, and wavelet domains to recognize the activities, but it is not clear how to determine the best features that can efficiently identify activities in less time. In this study, we aim to explore and elucidate the significance of the most relevant features in HAR, shedding light on their semantic meanings. We utilize different filter-based techniques like chi-square, correlation coefficient, and information gain, coupled with forward selection to find the top-ranked features. Additionally, we apply union and intersection operations on the resultant feature subsets of filtering techniques to obtain common and aggregated features. By considering these common relevant features, we observe a significant decrease in computational time by 97%, 95%, and 90%, with a reduction in memory usage by 10%, 25%, and 20% for benchmark datasets, i.e., UCI HAR and WISDM (phone accelerometer and gyroscope datasets) respectively, even though with a slight compromise in accuracy.

Although existing works in the literature highlight the monitoring, characterization, and analysis of both air and noise pollution, they mainly focus on the two environmental pollutants independently. In this paper, we develop a system framework that includes sensing and allows the processing of the combined impact of air and noise samples together to design micro-services. Few of the existing works that studied the combined effect of the two environmental stressors merely calculated the correlation values without further inferring contextual information from it. In contrast, our work aims to draw further inferences about the demographic/traffic/spatio-temporal aspect of a location and thus identifies the context in which the samples are collected. To achieve the goal, a system framework CoAN is developed under which we performed in-house data collection with approx. 820 km trail, covering approx. 10 km road segment in Durgapur, a sub-urban city in India. We used a commercially available ‘Flow’ device, and developed an android-based application, ‘AudREC’ for air and noise sampling, respectively. An unsupervised K-means algorithm has been used to segregate the combined samples into disjoint clusters for analysis. In addition, feature selection, model training, and cluster interpretation using the LIME model are performed to draw some inferences about the sample data space. Several supervised models, like Decision Tree, Random Forest, Logistic Regression, SVM, and Kernel-SVM are used for training the system. Results show that Logistic Regression performs best over others achieving 99% accuracy. Furthermore, as a micro-service, a healthier route recommendation system is designed to avoid pollution exposure by taking into account both air and noise pollution exposure volumes. A sample result shows that our recommended route gives almost 12% lesser pollution exposures as compared to all other available routes suggested by Google map with the same source and destination.

Noise pollution, a growing problem in urban cities, causes severe and long-term health problems. Vehicles on the road and ceaseless honking add to incessant vehicular assault and are the main reasons behind increasing noise pollution. However, analyzing and characterization of honking is not an easy task, and it varies with road traffic, vehicle types, speed, etc. In this paper, we propose a framework DeHonk that detects vehicular honks from traffic noise and analyzes various honking features to build a honking profile using only audio samples, which can be further used to infer the context of a particular location, design micro-services, etc. However, the existing methods used only fast fourier transform(FFT) or mel-frequency cepstral coefficients(MFCC) in honk identification, the accuracy lies between 40% to 60% in the presence of ambient noise. Similarly, we failed to obtain good accuracy in recognizing honks at noisy environments by applying existing sound classification models. In the proposed work, we have modeled the sound signal as a spectrogram image to detect vehicular honks. We have collected audio data by traveling 461km of road from various spatio-temporal locations in Durgapur, a sub-urban city in India. Furthermore, the generated spectrogram images are pre-processed, labeled, and fed into various deep learning models for training and testing. In this study, convolutional neural network(CNN), visual geometry group with a very deep convolutional network(VGGNets), efficient convolutional neural networks for mobile vision applications(MobileNet), residual neural network(ResNets), and Inception V3 models are used to train the system to identify honks and honk related features. Experiment results demonstrate that the ResNet152 model has achieved higher accuracy (97.69%) in honk detection as compared to other models. Furthermore, as a micro-service, a honk-aware route recommendation system is developed based on the derived honking features.

Increasing noise pollution levels insinuates the surging number of vehicular honking in the road traffic. Therefore, identifying the different vehicular honks and predicting the honking pattern can give a very good intuition about the noise pollution level in the environment. In this paper, a novel frame-work 'PredHonk' is developed that includes raw data sensing, honk identification and honk count forecasting to understand the pollution beforehand in order to avoid and minimize the effect of the pollution on daily human life. Although several researchers have done their research works on noise pollution monitoring and forecasting, but vehicular honk forecasting is not yet done. We collected the raw audio data from road traffic areas. After that, the audio files were converted into spectrogram images, which were fed into the ResNet152 model for honk identification. Identified honks were further used for predicting honk count using RNN, LSTM, Bi-LSTM, GRU models, and our proposed E-LSTM model. Results demonstrate that the proposed E-LSTM has shown good improvement in honk prediction.

In an urban area, the degree of ambient noise and air pollution play a vital role in determining the quality of human life. The impact of these two pollutants is increasing day-by-day due to rapid urbanization. Although, creating real-time pollution maps and forecasting of pollution levels have been studied extensively, the contextual, spatio-temporal correlation between air and noise pollution has not been investigated thoroughly. This correlation is important to identify the characteristics of an urban area. In this paper, we have highlighted some aspects that are useful to identify a context from different pollutant data. To collect the noise data, we have developed an android based application 'AudREC' that uses the inbuilt mobile micro-phone sensor. Moreover, for measuring air pollutants, we have used a ready-made 'Flow' device that senses PM2.5 and CO2, etc. The initial outdoor experiments show the feasibility of the platform for recognizing contexts from air and noise pollution information.

Intrusion detection system (IDS) is a software application which monitors the system, network activities, finds vulnerabilities if present, and protects digital data in a safe manner. An IDS monitors network traffic and data, features selection, analysis and action or detection, and also alert generation during life cycle. Firewall technique is one of the system-based protection techniques which is used to protect the private network from the public network. The areas where IDSs are used are in financial, healthcare, technical fields like MANET, cloud computing and its security, data mining. There are three types of intrusion detection systems—HIDS, NIDS, and APIDS. HIDS is based on sensors, where it can obtain data from operating system. HIDS can also tell attacker’s activity by analyzing network. NIDS is also based on network sensors. NIDSs can collect network information and can audit network attacks, while packet is moving across the network. APIDS works based on behavior and event of the protocol. IDS prevents various attacks based on OSI layer like DoS or DDoS attack, eavesdropping, spoofing, U2R, logon abuse, application-based. IDS system can be affected by the signal strength where in this paper the objective is to enhance detection rate of intrusion detection system based on Wi-fi optimization where we have tried to show that how some basic materials can affect intrusion detection system through Wi-fi signal strength.

Due to the exponential growth of NoSQL databases and in addition the circumstance of perusing humongous volumes of information, maximum applications switch RDBMS to NoSQL and pick it as information stockpiling framework. But we all know that RDBMS have several advantages which make it a popular platform across several applications over the decades. Therefore we view the standard problem of converting the RDBMS to NoSQL in reverse approach and we conceptualize a problem where NoSQL is converted back to a RDBMS based system. A generic framework is proposed in this paper so that different NoSQL databases could be converted to RDBMS. This approach is illustrated here using a case study on MongoDB and Neo4j. MongoDB is a document oriented database, fully unstructured and schemaless whereas Neo4j is a graph oriented database, fully unstructured and schemaless. This proves robustness of our proposed mechanism.